Stephen Morse

I see now that you have `JUNIT_REPORT_STACK`. So I guess the request here is just to have similar options for `JUNIT_REPORT_BODY` / `JUNIT_REPORT_FILE`

Ideally, I think it would be amazing to see a copy of the body of the test with the line that failed highlighted.

I talked to our dev-ops person, it seems that our development cluster has 3 brokers and a replication factor of 1. I copied this repo, put some debug lines in,...

> Is it producer? Simple or Group consumer? Can you share the code? I can't really share the code, but we do use a GroupConsumer and a Producer, although I...

Side note, you don't happen to have any plans to implement an equivalent of the [ConsumerRebalanceListener](https://kafka.apache.org/090/javadoc/org/apache/kafka/clients/consumer/ConsumerRebalanceListener.html), do you? I'm guessing not. The dev-ops person who manages our Kafka clusters thinks...

> If I correctly understand those messages HEX dumps, they are ListGroup requests. Are you using GroupAdmin and its listGroups method? Yes, we use that as part of our health...

> Are there messages such as Rejoining group on RebalanceInProgress or others? No, but there may be a logging configuration problem.

> Can you stop using listGroups request and/or GroupAdmin and see if it resolves duplicate messages problem? I removed the call to listGroups but kept the initialization of the the...

@oleksiyk The problem went away when I started using `describeGroup`, though. Wouldn't that call have the same metadata issue if GroupAdmins weren't refreshing their metadata?

Switching from `listGroups` to `describeGroup` has gotten rid of the `cannot read property 'resolve' of undefined` errors, but has surfaced a few protocol errors: ``` RangeError: Trying to access beyond...